A novel method for estimating subresolution porosity from CT images and its application to homogeneity evaluation of porous media

We propose a new method, i.e., the statistical phase fraction (SPF) method, to estimate the total porosity and spatial distribution of local porosities from subresolution pore-dominated X-ray microtomography images of porous materials. The SPF method assumes that a voxel in a CT image is composed of either a single or a maximum of three pure phases of matter (solid, liquid and air). Gaussian function (GF) fitting is conducted on the basis that the summation of the area of each GF curve is equal to the total area covered by the CT histogram. The volume fraction of each phase corresponding to each GF is calculated based on the mean value of the GF, the area of the GF, and the CT numbers for pure phases. The SPF method is verified on three different types of components containing only air and solid phases, i.e., alumina ceramic and two sintered lunar regolith simulants with relatively homogenous and inhomogeneous microstructures. The estimated porosities of a total of 15 specimens (the total porosity ranges from 0 to 51%) via the SPF method show an average error of 3.11% compared with the ground truth. Spatial distribution of local porosities in the defined representative element volume is investigated for homogeneity evaluation. Results show that the local porosity inhomogeneity in the sintered FJS-1 specimens is more prominent than that in the sintered KLS-1 specimens.

Automated summary

A new method called statistical phase fraction (SPF) method is proposed to estimate the total porosity and spatial distribution of local porosities in porous materials. The method is validated on different components and shows high accuracy when compared with the ground truth.